rxrpc: Fix oops in tracepoint
authorDavid Howells <dhowells@redhat.com>
Tue, 2 Jul 2019 15:04:19 +0000 (16:04 +0100)
committerDavid S. Miller <davem@davemloft.net>
Tue, 2 Jul 2019 22:29:57 +0000 (15:29 -0700)
commit99f0eae653b2db64917d0b58099eb51e300b311d
tree7e697394bbf9030ce6e756381a83bd0e850cf754
parent78226f6eaac80bf30256a33a4926c194ceefdf36
rxrpc: Fix oops in tracepoint

If the rxrpc_eproto tracepoint is enabled, an oops will be cause by the
trace line that rxrpc_extract_header() tries to emit when a protocol error
occurs (typically because the packet is short) because the call argument is
NULL.

Fix this by using ?: to assume 0 as the debug_id if call is NULL.

This can then be induced by:

echo -e '\0\0\0\0\0\0\0\0' | ncat -4u --send-only <addr> 20001

where addr has the following program running on it:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/socket.h>
#include <arpa/inet.h>
#include <linux/rxrpc.h>
int main(void)
{
struct sockaddr_rxrpc srx;
int fd;
memset(&srx, 0, sizeof(srx));
srx.srx_family = AF_RXRPC;
srx.srx_service = 0;
srx.transport_type = AF_INET;
srx.transport_len = sizeof(srx.transport.sin);
srx.transport.sin.sin_family = AF_INET;
srx.transport.sin.sin_port = htons(0x4e21);
fd = socket(AF_RXRPC, SOCK_DGRAM, AF_INET6);
bind(fd, (struct sockaddr *)&srx, sizeof(srx));
sleep(20);
return 0;
}

It results in the following oops.

BUG: kernel NULL pointer dereference, address: 0000000000000340
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
...
RIP: 0010:trace_event_raw_event_rxrpc_rx_eproto+0x47/0xac
...
Call Trace:
 <IRQ>
 rxrpc_extract_header+0x86/0x171
 ? rcu_read_lock_sched_held+0x5d/0x63
 ? rxrpc_new_skb+0xd4/0x109
 rxrpc_input_packet+0xef/0x14fc
 ? rxrpc_input_data+0x986/0x986
 udp_queue_rcv_one_skb+0xbf/0x3d0
 udp_unicast_rcv_skb.isra.8+0x64/0x71
 ip_protocol_deliver_rcu+0xe4/0x1b4
 ip_local_deliver+0xf0/0x154
 __netif_receive_skb_one_core+0x50/0x6c
 netif_receive_skb_internal+0x26b/0x2e9
 napi_gro_receive+0xf8/0x1da
 rtl8169_poll+0x303/0x4c4
 net_rx_action+0x10e/0x333
 __do_softirq+0x1a5/0x38f
 irq_exit+0x54/0xc4
 do_IRQ+0xda/0xf8
 common_interrupt+0xf/0xf
 </IRQ>
 ...
 ? cpuidle_enter_state+0x23c/0x34d
 cpuidle_enter+0x2a/0x36
 do_idle+0x163/0x1ea
 cpu_startup_entry+0x1d/0x1f
 start_secondary+0x157/0x172
 secondary_startup_64+0xa4/0xb0

Fixes: a25e21f0bcd2 ("rxrpc, afs: Use debug_ids rather than pointers in traces")
Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/trace/events/rxrpc.h